The Exchanges

Q Um, here is, and to wrap up the AI segment, if you're open AI, and you see Grok, You're Sam Altman. You're playing with this thing. What goes through your head?

A Well, so Dave and I did this cost analysis over lunch, because that's what we do, of how much money has gone into each to build what seems like very competitive platforms, LLMs. Um, and so, twelve billion has gone into Grok, and, and OpenAI, you have to sort of discount the Microsoft dollars, because a lot of it's for compute, but an estimated twenty billion. Into open AI. So, you know, almost double what Grok has used to build a very similar, if not better performing, LLM. And so I think, I think Sam Altman and the open AI team should re like really think, you know, and this was what happened with deep seek, right? It was like, how could we be undercut by costs so much? And so I think that's a real outcome of this.

AI assessment note: “how could we be undercut by costs so much?”

Q a war. And so as a result, like, you just want to be like, no, like this is, we're now fighters. AI's are new weapons. And like, we're fighting over the same territory. We're just doing it with new tools. And I think that's just like a different mentality where like, in a fight, you're like, I'm fighting Bill. I'm fighting Cloudflare versus like, no, it's all new, you know?

A It, it really depends on who. For, for startups, we're, we're hardly ever, for AI startups, we're hardly ever fighting Cloudflare. So for example, uh, uh, just to use the example from the election, Polymarket, it was always, it was born on, in Vercel. Uh, for Washington Post, we're fighting Akamai. And for Minnesota Star Tribune, we fought on-prem. And they moved off of the legacy. Into the new homestead land of Vercel. So it depends on who you're talking to. I think for enterprise, you're absolutely right. Like when you get into like deep enterprise, everyone's fighting for your workload and like the big banks and, and, and, and the big insurance companies, they have 20 vendors courting them at any given point and it's more of a knife fight.

AI assessment note: “It, it really depends on who. For, for startups, we're, we're hardly ever”

Q Well, and I saw like your, your growth rate is also crazy, right? Like, is this all because of the agent swarm?

A Yeah, it's because it's that same intuition, right? That like humans, uh, Vercel's growth was capped by how many humans exist and how many times they deploy by submitting PRs. If now coding agents start taking that job of writing the code and deploying it and testing it and submitting PRs, you know, GitHub also published their data and it's insane. Now, all of a sudden, our infrastructure has like a hundred times the demand. Our signups are growing by 50% month over month. My high school friend, who's never written a line of code in their life, came to her office yesterday, and he's like, by the way, I, I've been following you, obviously, and what you've been doing for years, and in Argentina, there's always news and whatever, but I never understood what Vercel was, but now Claude introduced me to Vercel. And so there's this new, like agentic driven interaction, and I fully expect agents to be signing up for services directly, zero human intervention soon, making payments on behalf of people imminently, you know, getting credit cards and budgets and solving entire problems for, for people where the service provider becomes an, a vendor of the agent, not a vendor of the human.

AI assessment note: “Yeah, it's because it's that same intuition, right?”

Q And to put this in context, that tweet was what, less than 18 months ago? And remind people what the tweet was.

A Yeah, it was this idea that generating software, it was in competition now with Googling for software, but now generating software is not just in competition, it's going to win. It means that everyone can produce their own software for personal use or for enterprise use, because why would you go through the hard work of procuring a new vendor if you've already procured Vercel, you plug it into Codex or Cloud Code, and now you generated this software that you thought you were going to procure. Like, this is literally happening as we speak. What makes Vercel really good for this hosting solution for generative software? One of the things that we bet on was serverless. Serverless has this property of scale to zero. A lot of software will be so spontaneously created and used that it might just get intense use for five minutes, for two weeks, maybe a year. Maybe it's just for me. Maybe it's for a medium-sized group. Maybe it takes off on X. Like literally every day there is a new, a generative personal app that takes over the world on X. There's one called the world monitor, which is vibe coded Palantir. It literally became one of the top 100 host names on Vercel for a week because everyone in the world was talking about it. And so.

AI assessment note: “it was this idea that generating software, it was in competition now with Googling”

Q Do we want to give the, uh, prognosis, Jess?

A We'll give the prognosis. Sam did not break his shoulder, his rib, or his collarbone. So he's in significant pain, but it's not broken. So anyway, dear listeners, thank you for that. Now we are ready. And Sam may or may not show up, but it will have nothing to do With his, um, perfectly intact bones. So, okay, let's start with OpenClaw. So the information reported this week that, uh, the founder of OpenClaw was going over to OpenAI. The announcement was quickly followed by OpenAI announcing that the news was true. And interestingly, that a new foundation for this red hot open source project that we've been talking about nonstop on this pod Called Open Claw. Uh, this foundation would be established, and who would be on this board of said foundation? Well, none other than Dave Morin. So, we are honored to be in the presence of Open Claw Foundation Board greatness. Dave, what can you tell us about this exciting project?

AI assessment note: “We'll give the prognosis. Sam did not break his shoulder, his rib, or his collarbone.”

Q Um, yeah. What about for Ben and Nick? Okay, so Doge, what is the sentiment on the parade of tech billionaires who were currying Trump's favor at the moment? I mean, do you see that from your perspective and Do DC insiders have the same attitude as like, oh, how cute, we still run the show?

A I think the, I mean, the people I, I mean, I think, I mean, I mean, maybe this is sort of my view and the view of some people I talk to, that, that Satya looks really, really smart for staying out of that photo. And that, and that, but that there was, and that there's a sense when you talk to some of the people around some of these, um, You know, these, these tech billionaires that, like, they're like, oh, we finally solved politics. We finally figured it out. Like, we just go all in on Trump, then we never have to worry about politics, and it's going to be great. And then what is it, like, one of the, you know, if the Democrats take the House, we're talking in 20 months, there will be a corruption committee convened, which will have Mark Zuckerberg, Tim Cook, Sundar Pichai, and whoever talking in great detail, but every penny that was directed, and, you know, and I think they'll want to scalp. Like, I think the Democrats will really want to destroy one of those people. And I, like, and I think Elon is probably too big to fail for the US government. So, like, if I were Mark, I'd be really worried. I think he's, like, wildly overexposed, and did, did not manage to make friends with the Republicans. Like, they think this whole thing is funny, obviously. It is funny. It's impossible not to think it's funny. And then the Democrats really are out for blood now. And so I, I actually …

AI assessment note: “business leaders in the United States typically do not do stuff like that”

Q in our universe say like for us to invest, we, we have a lot of stuff coming through our network too. And we also are out there, you know, hunting and getting like blind intros to founders. Did you come across that strategy of network only just while you've been at the university of Michigan, or is that something you saw work and felt work and earlier in your career?

A I think it's sort of sunk in over time. That's what I, Learned when I was at Princeton working with Andy Golden, and basically the idea there was that we want to be in business with really great people, great groups, and that with a small team trying to cover the world, and with the networks that, that Princeton has, and with later when I was at Wash U and now at Michigan, there, there's some really phenomenal networks. It's really not playing to our strengths to go out and meet everybody who's got a fund and try to be a picker. Being a picker is really, really hard. It's hard in any, in any sense, right? If you want to be a stock picker, you know, these things are done in isolation. And so what we're doing is just leveraging the things that we have that are basically in our own portfolio, which are connectivity with people, connectivity with information that other people just don't have, and then doing our best with that.

AI assessment note: “That's what I, Learned when I was at Princeton working with Andy Golden”

Q Have big audiences in this realm. And they've just kind of said, you know, like this has gotten to be too much. Like I can't, we can't produce the content at the speed that I need to, to keep up with the algorithm. And I don't know what, there seems to be a lot to unpack there. And I'm just curious, Casey, like what, what you think is going on?

A Yeah. I mean, like Tom Scott, who is a fantastic YouTuber and he's kind of in that space, very big channel, very thoughtful guy. He just announced that he's, he's sort of giving up YouTube. You know, I think it's, I don't think it's any one reason, but I do think it's a confluence of factors, and I think I'm, I'm victim of some of those factors as well. I attribute my ability to navigate them better than some to the fact that I was kind of mature in the space before I embraced YouTube. You know, I had a TV show on HBO that I starred in 10 years before I really found success on YouTube. I was, I, I was mature when I entered the space. But, you know, some, for a lot of these creators, their, their passion or their reason why they're there Typically, like, is not the fame, not the celebrity of it. It's the desire to create. It's the desire to share meaningful stuff. And there's all these secondary and tertiary aspects and things that come along with being your creator that are wildly burdensome. And some are technical, like the fact that, you know, maintaining, like, it's undeniable. If you maintain consistent uploads, you will grow. And if you don't, you will not grow. I do not upload consistently. Like my, my subscriber count, which would, used to grow at, you know, um, hundreds of thousands a month. It has been stagnant. Like, since I stopped, I still do great views, but it's b…

AI assessment note: “there's all these secondary and tertiary aspects and things that come along with being your creator that are wildly burdensome”

Q like right now I feel like we're basically, and this is partially the cost of the models and the cost of you can compute for them. We're asking every question we can come up with to a really, really expensive PhD. And in the future, I think we'll be a lot smarter about who, what level of intelligence we need for different questions. How do you see that competitively playing?

A I might disagree. I think what will happen is the expensive PhD will become dirt cheap. My bet is, a year from now, it'll cost one-fifth to one-tenth of what it does today, and my advice to all our startups is ignore the cost of compute, because any assumption you make, any dollar you spend on optimizing your software will be worthless within the year, and so forget about it. Rely on competition in the marketplace between thought and Gemini and OpenAI. To, to, to reduce that cost to a point where it doesn't matter. And below a certain level, it doesn't matter. You're paying a certain amount for your iPhone service monthly. If it's 10% or less of that, it doesn't matter. Yeah.

AI assessment note: “I might disagree. I think what will happen is the expensive PhD will become dirt cheap.”

Q We need to know at least one thing that Suds is the most interested in right now. Like, what's your favorite company you're working on, man?

A I just incubated a CI company, uh, with Kleiner and Benchmark. That one is going to do well. Basically, underlying problem, every human at any of these companies, like engineers, they were all writing like one X code. Now you have 50 X more code to ship the exact same feature. It's breaking all the Pipelines and like traditional ways of testing, merging, shipping code. And now you basically like drive up your CI costs, you drive up costs associated, and like the speed goes down a lot associated with like shipping code. And so the whole problem statement's like, how do you solve for that? And so you start with CI, uh, that's like the first part of it. And then from there, it's like, if you really integrate yourself at this.

AI assessment note: “I just incubated a CI company, uh, with Kleiner and Benchmark.”

Q How different was it in the Facebook days? Because you had sort of the same dynamic.

A I think, first of all, it's kind of funny because by historical standards, then it was an incredibly meteoric rise. Facebook's valuation grew unbelievably slowly compared to the labs, right? And There were obviously generations. Like I would argue I was Peter, you were kind of generation one and a half. Maybe I was generation two at Facebook. There's now been generation three and generation four, but there's not 20 generations, right? There was like a set of people who worked for four years, did great work, and then passed the torch. And that's happened like a few times. And actually the core leadership team at Facebook has been unbelievably stable now for like a decade, right? Like Which is a very different dynamic.

AI assessment note: “Facebook's valuation grew unbelievably slowly compared to the labs”

Q Well, Peter, can I push you on that? That is the maybe URMI narrative, but there was a study I saw recently that said that AI is so hated, it's actually hated more than Jeffrey Epstein. By the general public, right? It's like, it is absolute, it is the worst brand like you could possibly imagine. No one trusts it.

A I'm not disputing that. What I'm saying is that that barometer is going to shift over time. I was just on a call with a bunch of creatives, and maybe Scott, I'd love to hear your thoughts on this because you're in the thick of it, who, you know, they're writers who just started saying like how awesome it's been for them to build portfolios for their, you know, friends with cloud code that they were unable to do before. That's like not a conversation I would have had with them Six months ago. I'm not saying where the trust is right now, Sam. I'm saying that that shifts as people get more comfortable with it, and it's just a matter of time.

AI assessment note: “I'm not disputing that. What I'm saying is that that barometer is going to shift”

Q like not have that happen. I want to pivot one or two more times before I lose you guys, and I really appreciate you, you joining for this fun conversation. OpenAI Exec Exodus. Brad Lightcap out. Kevin Weill, who we used to work with, out. Peter, you've been out for, like, years, so, like, you were just, you're a forebearer on this. Like, why and, like, what, what's going on?

A My take on this is it's actually a feature, not a bug. It's a feature of having insanely great talent. I think Sam is N of one at recruiting with really ambitious, talented people. And in a time when you can build anything, you should expect the churn. These are people who want to go and kind of build something really, really insanely great. And that's not a talent problem. That's just kind of a natural output of what he's optimized for. I mean, you're seeing this everywhere. And also the other thing is that it's not just having the models. You have to point the models with Sense of obsession at a problem. And, you know, like something like periodic labs, uh, Liam left, uh, open AI. We were the first check investors into that. They're doing material science. It requires a different level of obsession on building a wet lab that just doesn't exist at a bigger companies, right? So the companies out there like applied compute, core automation, other new labs, there's just that there's a point of view that you want to go chase. And if you're ambitious, you're going to go out and chase that. That's largely what I see from this. You have to point the models and it's a great time to build. And so that's what everyone is really doing.

AI assessment note: “it's actually a feature, not a bug. It's a feature of having insanely great talent.”

Q Fair enough. Scott, we didn't really get into proof of craft, providence, watermarks. You've been thinking about this stuff forever. Anthropic just is now watermarking text as of today. Like, there's a lot of stuff going on. I mean, I, I assume you are still deeply on this train. Can you give us a quick update on that?

A The CEPA, it's the open protocol that Kind of everyone signed on to for these content credentials to be added to assets as they're generated or edited, by the way, for that matter, it has been widely spread. The problem is on the consumer experience side. These companies like Instagram, you know, and products that are surfacing media for them, it's binary. It's like, if it has a CGPA in it that says any AI was used, they reported as being made with AI. And then consumers are like, what the hell? I only use AI to like remove a blemish from my face. Like, why are you labeling me as a slot maker? And, and so there's, we're in this like moment where there's a disconnect between the underlying efforts, which I think are net net good to like help people be able to discern the provenance if they want to. And the consumer interfaces that don't know how to surface this or merchandise this to the end user. So I feel like that's where we're at right now. I don't think it's going to go away. But I also think we're all gonna be gradually inoculated, and I think, honestly, like, fake stuff is really helpful for society right now because it inoculates us and makes us realize we can't trust what we see anymore.

AI assessment note: “So I feel like that's where we're at right now.”

Q Okay. Well, speaking of building things, Dave, you were at the Microsoft Build Conference this week. What happened there? Anything of note?

A It's a continuation of the same message that we're hearing from all of these big developer conferences. Although I guess it's differs slightly from the Google one in that Microsoft is very focused on agents and the enterprise. And, you know, we did a big integration effort with Microsoft over the last month or two. We announced that Open Claw is fully integrated with Windows now and did a bunch of Really great security work around giving it controls. And, you know, if you want to deploy open cloud into the enterprise, now you can using all of the windows and Azure tools, and you can have full control over it and, you know, set up all of your observability policies, admin policies, all that kind of stuff. It's the stuff people are asking for. They also really kind of cool integrated it like directly into windows and they have a new long running agent called scout. That's based on open Built on Openclaw. And so that's been pretty neat to see, but you know, that's the same story that a lot of folks are chasing right now and kind of in the story of this half of the year. So, um, it's cool. It's cool to be on stage with Microsoft.

AI assessment note: “Microsoft is very focused on agents and the enterprise.”

Q There's, you know, a Google IO, Microsoft build. We've got Apple next week. We've got the Dell world event happening. Like, like, does a lot of us feel like feature parody? Like everyone's trying to get to the same Infrastructure when it comes to AI and agents in the big tech land?

A No, I, I mean, I do think that this is the era of figuring out What agents are going to be. I keep saying this, but I think it's kind of 1997. I always say it's like either 1983 or 1997, depending on which part of the technology canon you want to look at. But we're kind of in that era where websites were just invented. Some of the early languages for programming them are around some of the early servers for building them are around, but the, you know, real, really cool stuff that you're going to be able to build with this has not yet emerged and it's Pretty messy. And agents, like the narrative between what agents can do and what people think they can do could not be wider. That's like one of my other narrative fascinations right now, Sam, which is that it's like anybody building this stuff knows that like they cannot do what a lot of people are claiming that they can do, i.e. they cannot be an employee of your business. Like they do not learn every day. They have really poor memory. There's a lot of stuff to figure out.

AI assessment note: “No, I, I mean, I do think that this is the era of figuring out”

Q these things. And I'm just like, what you just said, the EQ component of this is, it's like so critical because it's a human, it's one of the most human jobs, right? I mean, Dave, David, to reflect, you saw that all happening. Obviously we talked about it at offline and I mean, we have, we've had a lot of these other incidents with our own founders. Wouldn't you say?

A Yeah. I mean, this is a deeply human business, right? Being a founder is incredibly lonely. Um, it's very difficult. Those of us that have been through it know this. And I think that there are any number of things, I guess maybe the worst of which is your home and entire community burning down on the one end, but there's, there's just a, an enormous number of things that can go wrong that are all human, right? And I think that a big part of the venture job is to actually show up as a human and to actually be there, help founders, whether it's helping them through their own Traps in their own thinking, uh, their own sort of internal struggles to their, you know, social and community struggles to these types of things, right? Like it will kill the company just the same if the human breaks down. Right.

AI assessment note: “Yeah. I mean, this is a deeply human business, right?”

Q whatever, it's kind of like, The background radiation of the universe, right? Which is like, but there are cash flows associated with it, right? Like things are like, so talk to me about like, you know, it's not all just crypto, right? There's like these different narratives of like where crypto derives value and like what the system looks like. And how do you think about those things as different?

A So the Bitcoin is so abstract that I cannot rationally value it. So I don't consider it an engineering problem. There's no amount of work I can put in all my, you know, Illinois computer science education isn't going to help me here. So Solana is not Bitcoin or is trying to do something that is, I think, easier to understand. It's just a dumb information message bus. People need to send messages that have Value associated with them. Like, I want to sell a coin, whether it's Bitcoin for USDC or something like that. I want to make a trade. That's a bit of information. So you post it, and then that offer is valuable for whoever wants to be the first to match against it. And because only one person can match against any offer, there is now an economic opportunity to be first, and you'll pay a priority fee to be first. So the more offers there are offered on this message bus, the more it makes, the more money it makes. So as long as we get more assets traded on there, more coins, doesn't matter what they are. Like the message bus doesn't even know that these things are, are coins or whatever. It's just bits coming in and out and people set a priority of how fast they wanted to land and they pay more for a higher priority. So it's kind of like if

AI assessment note: “Bitcoin is so abstract that I cannot rationally value it... Solana is not Bitcoin”

Q Apple, big announcement that we have all been kind of waiting for, the unveiling of Apple intelligence. Uh, that's a great place to start this week. And of course, A big part of that was Apple's partnership with OpenAI, which Vinod was the first venture capital investor in way back when. So, so I got to ask Vinod, what, what did you make of all the news out of Apple?

A Well, first, I haven't caught up with all the news. That was reserved for the weekend. But Apple needed to do something. Siri was starting to go from a reputation to worse, if that was possible. But more importantly, They've demonstrated something that's really important. Computers interface with you directly. You don't have to futz around to get stuff from computers. Computers get you stuff. And I'd done a blog post in this, on this topic a couple of months ago in the information, this idea that computers will learn humans instead of humans needing to learn computers. And I think we were seeing the beginnings of that. That was exciting. Obviously from an open AI perspective, it clearly establishes where open AI is in the competition for the best LLMs. Obviously many people wanted that business. And many things were evaluated, and so it's really rewarding to see where Apple thinks the best AI is, and, uh, and I'm sure they considered where the best AI will be a year or two from now. Uh, so in many ways, validation for OpenAI and a really big milestone in how humans interface with machines.

AI assessment note: “in many ways, validation for OpenAI and a really big milestone”

Q Is it just what you're saying that these businesses are, the margins are so razor thin that unless you go and do luxury version of it, you're not going to be able to produce the margin. Like, is it as simple as that?

A I think it is as simple as that. I think that you can kind of bifurcate the market and you can say, You can say, Hey, I want to go super expensive, 15, 20, 30, 40,000 dollars a year per person. Don't venture back it. Um, and say, I'm going to build a nice, like slow growth, reasonable business. But let me be clear. You're not affecting like, you know, the, the common man, right? This is not for, for everybody. Or you can say, Hey, I want to push towards massive scale, which is kind of the, the forward attempt of, uh, I'm going to be very low price. I'm going to try and be affordable enough that everybody in society can get it. And our whole goal, uh, was that because we're using technology as opposed to kind of an enormous amount of labor that we're able to kind of just reduce that price more and more over time till it becomes the commodity that everyone has. I think both of those strategies are reasonable strategies. Um, it just depends on what's your goal. Like, why'd you start the company, right?

AI assessment note: “I think it is as simple as that. I think that you can kind of bifurcate”

Q Well, no, I mean, I think that's what your choice to do this, this fast. I mean, I'm, I'm, I'm here pushing you on these things because it's worth, it's worth thinking about, like, do you think you should have?

A I would not have done that. Uh, and, and I will not. And again, somebody else should go try that if they think it's a good idea. I, I will not go do that. It's like, it's like Larry Page sitting down with Sergey and being like, hey, we want to invent Google. Step one, let's learn the Dewey Decimal System and visit the Library of Congress. It's like, what are you talking about? No, like, like, you know, you're not gonna, you're not gonna disrupt it by trying to, uh, to create the existing system. Again, you could make improvements and you can have impact. I, I don't want to say you can't, but I don't think you're gonna create that generational kind of new company.

AI assessment note: “I would not have done that. Uh, and, and I will not.”

Q mean, there's something incredible, like, just think about this. Like, we just burned, you know, not quite half a billion dollars on an idea, and every, no, I shouldn't say everybody, but a lot of people's reaction is, What are you doing next? Let's do it again. And like, what, what sort of specialized culture did we create in Silicon Valley where this is reality? I mean, this is absurd.

A Well, to give you credit, like, I think the quote that, that Dave and I use a lot is like a lot of entrepreneurs can like lift off the plane, but not all entrepreneurs can land the plane and landing the plane could be an exit. It could be returning partial money back, or it could just be like, Following through to shut down the company in the right way. Just finish the job. Don't abandon the ship. Like, and literally the credibility of when an entrepreneur can just do that with, you know, fight the good fight, have a big idea, like have a great culture in the company and like really go for it. Then land the plane when it doesn't work. That is all the credibility you need to get the investors that want to invest in you the next time.

AI assessment note: “That is all the credibility you need to get the investors that want to invest”

Q is really more apt to how I think a lot of these people do think about it. And, you know, many of them have told me flat out, you know, you, who are the nodes at your company who do X, Y, and Z? And I, I always thought that was fascinating because it, Turns a lot of it on its head. But Britt, what's your reaction to all this?

A Okay. Two things. One, now that I'm up to speed and read the essay, Paul Graham does talk about the idea that a lot of founders are gaslit by their own VCs, right? And the problem with that is most VCs haven't been founders themselves. And I think that's becoming more and more important. And obviously all four of us on this pod are former founders, current founders. And We know what it takes to be a founder. And I think that if you are a founder out there, it's not only good for you to find VCs that have been founders and understand what it's like to be in that position, but I also think like you should consider your board composition in a way where even your non-investors on the board might have more empathy for you as a founder or understand what it's like to be in that chair. I think there's just, there's a lot more that goes into the full understanding of that role when you have those people on the sidelines Like championing you.

AI assessment note: “Paul Graham does talk about the idea that a lot of founders are gaslit”

Q that, and you know, as the author of probably one of the best handbooks for how to grow a startup that exists, how do you advise founders to think about, you know, it's going to be 10 years. We, you might end up failing once. You might end up failing twice before you figure out what works. Like, how do you advise people to think about a 10 year journey?

A Yeah, I think, um, I think there's a few different ways to think about it. I think on the downside, in some sense, I think if things aren't working, I think because money has been so available, people have kept their companies going longer than they should. And in a lot of circumstances more recently, where sometimes the best thing you can do is just kind of quit and restart and go do something different. And the way I kind of think about it is, you know, as a founder, if you're able to go and start a company, those are probably the best, most productive years of your life in terms of ability to take risks. And therefore, if the thing isn't working, move quickly onto the next thing, right? And that goes counter to the advice of grind forever and an iteration 73.

AI assessment note: “if the thing isn't working, move quickly onto the next thing”

Q like, right now I feel like we're basically, and this is partially the cost of the models and the cost of you can compute for them. We're asking every question we can come up with to a really, really expensive PhD. And in the future, I think we'll be a lot smarter about who, what level of intelligence we need for different questions. How do you see that competitively playing?

A I might disagree. I think what will happen is the expensive PhD will become dirt cheap. My bet is a year from now, it'll cost one-fifth to one-tenth of what it does today. And my advice to all our startups is ignore the cost of compute. Because any assumption you make, any dollar you spend on optimizing your software will be worthless within the year. And so forget about it. Rely on competition in the marketplace between thought and Gemini and open AI. Uh, to, to, to reduce that cost to a point where it doesn't matter and below a certain level, it doesn't matter. You're paying a certain amount for your iPhone service monthly. If it's a 10% or less of that, it doesn't matter.

AI assessment note: “I might disagree. I think what will happen is the expensive PhD will become dirt cheap.”

Q And w what do you think of AI as a category? I mean, you guys are big and open AI.

A Yeah. I mean, it's not a category. I mean, I, I, I, I, I've said this stat a couple of times, but you know, when I first started doing venture capital, everything was social, local, mobile. It was like over 60% of the companies that I looked at, yeah, looked at in the first year were, were of that category, and it was like, ah, everybody was doing it, right? Now, every single, for a while, I think the pendulum was already being on the shift, but for a while, it was close to 100% Of companies would say AI or LLM by, you know, minute two of the pitch or page one of the deck. And I think that that's changed. It's now just, uh, you know, a tool like it's like saying the internet and people have realized that. So I think people are less and less trying to overload that term, which is really good. Um, but, so I think it was just, that was a challenging few months there where every single company was just saying, oh, we're an AI company. It's like, no. And so.

AI assessment note: “Yeah. I mean, it's not a category.”

Q So you believe, like, you believe that there's, you know, is it stronger now than it was last year, Dave?

A Uh, Yes, there's more people talking about it. I think that the, um, you know, one of the things I had heard a few years ago was that the focus on this is not just in the air, but it's in the water. And the fact that when you read the appropriations documents, uh, that have been moving through the Senate and et cetera, they changed the name of this, uh, committee that they formed to be the Underwater and Aerial Phenomena Study Committee. Um, I had heard that it was something, you know, that I had heard a while ago, but the fact that these things are actually starting to move into these bills and into, like, very serious procedural machinations of the government says to me that there's more here than, than we might know. Um, but, you know, I'm a data-driven guy, so I want to see the data.

AI assessment note: “Yes, there's more people talking about it.”

Q Just quickly. Like when you think about how you do that, is that like more cash in the bank?

A Is it was so, so Uh, everybody here presumably knows Mujal Shah, who's a good friend of mine who, um, has a startup called Hippocratic. That's really exciting. Before this, he did a startup that, that didn't work out. And so I asked him like, well, how are you approaching Hippocratic? And he, I free, I'm not going to use the right terms, but he's like, we basically have this go cold plan. Anything goes wrong. We can scale down our burn and ensure that like with the contracts we're signing, we're never over committed. We can always scale down to a point where we've got, and I forget his number, 36 months of cash in the bank. I actually thought that was a really intelligent way to think about it.

AI assessment note: “scale down to a point where we've got... 36 months of cash in the bank”

Q you're talking about billions and billions and billions of fuck around dollars, right? Being thrown around. And how do you feel about that as an investor, right? And as a capitalist, when you think about, yes, I mean, I agree. In enterprise value from some of this fuck around stuff, but the cost is going to be a trillion dollars of capital sunk to create that three hundred billion dollars.

A Yeah, it's, it's a really interesting question, because if you look at some of the numbers that are being quoted in terms of revenue, the revenue basis of these things is really high too. I'll come back to that in a second. I think in prior waves, we also saw mass scale investment. I mean, crypto was huge sums of money invested in ICOs and You know, even five, six, seven years ago, bad meme coins and other things, um, the internet on a relative basis to prior venture capital scale was massive investment in the internet, including public markets because companies used to go public within three, four years. And so the public market investors were really the flood of capital that happened there versus private. So now we just have these long private things. But if you look at it, for example, Azure in their last quarter announced a twenty five billion dollar quarter of which a five percent lift came from AI. So they added something like a billion, a billion and a half per quarter in new revenue for what they considered AI services on their cloud. And that's one cloud provider, right? So if you extrapolate that forward at say a five billion dollar run rate, assuming no new growth just for Azure. And if you assume, you know, AWS and GCP are seeing something reasonably similar, you're talking about 10 to fifteen billion dollars in revenue today that's growing really rapidly for AI stu…

AI assessment note: “the revenue basis of these things is really high too”

Q Like take us to 20, 21 for a second. Like, like, so when you were, when you saw those big check, those big valuations, like, I mean, you're a smart man, right? You know that that would wrap up expectations, but what was it about this moment that made you want to swing for the fence?

A Well, I wanted to swing from the fence from the beginning. Um, so just taking a step back, like I, you know, I've, I've done, uh, I've done enough things in my career where it's not like I'm like sitting around going, oh, you know, getting a single or a double, is that interesting? You know, at some point, this is your life more than this is your career. And I just want to see if I can, you know, improve humanity for the better. That's really important for me. And I know that healthcare is easily one of the largest problems on this planet. It literally affects every single one of us. Unless, you know, somebody who's not going to die, maybe Brian Johnson, it doesn't and everybody else, we, we, we need a healthcare solution. And so I said, okay, let's go for it. Now we always took a stance that like the, the approaches that everybody has in the world of healthcare are incredibly incremental right now. Um, you can't look, you know, maybe, maybe I'm biased, maybe I'm wrong, but if you go look at every single healthcare company out there, you don't go, wait a minute, if they're successful, everything changes. And what I wanted was the company where it's like that generational company where you say, you know what? We've kind of just put a pin in, in that, in that kind of healthcare problem. And we no longer really have to think about it anymore. I think a lot about like, what did Goo…

AI assessment note: “Well, I wanted to swing from the fence from the beginning.”